Virus-based vectors such as adenovirus and adenovirus-associated virus (AAV) have been widely used in gene therapy applications due to their high efficiency of transduction into a variety of cells in vivo. Indeed, several impressive results using viral vector-based gene therapy have been reported in humans (Ashtari et al., 2011; Fischer et al., 2011). However, one major barrier to successful gene therapy with these viral vectors is the host immune responses to both viral vectors and the transgenes (Huang and Yang, 2009). Our understanding of the mechanisms of immune responses to viral vectors has greatly improved over the past two decades, and as a result, strategies have been developed to manipulate various immune pathways in order to regulate the immune responses to viral vectors. One promising therapeutic strategy is targeting the co-stimulatory pathways to modulate the function of T and B cells.

The original two-signal hypothesis (Lafferty et al., 1974, 1980) proposes that full T cell activation requires two signals: antigen-specific T cell recognition (signal 1) provided by the engagement of the TCR–CD3 complex with a processed antigenic peptide bound to the major histocompatibility complex (MHC) molecule on the surface of an antigen presenting cell (APC), and an antigen-non-specific co-stimulatory signal (signal 2) provided by the interactions between cell surface molecules on the APC and the T cell. Indeed, early work by Jenkins and Schwartz showed that in the absence of the co-stimulatory signal, T cells are rendered anergic, a state marked by the inability of T cells to respond to subsequent antigenic stimulation, leading to a model that co-stimulation plays a critical role in controlling the fate of T cell responses: activation or anergy (Schwartz et al., 1989; Jenkins et al., 1990). Since then, these seminal studies have stimulated tremendous growth of the co-stimulation field. In addition to signals 1 and 2, we now appreciate that other signals also participate in determining the fate of T cell activation: some of these signals are stimulatory and others are inhibitory (Sharpe, 2009). Furthermore, some other signals are required to promote the differentiation into different T helper cell subsets (Pipkin and Rao, 2009; Simpson et al., 2010).

Because of the importance of co-stimulation in determining the outcome of the immune response, manipulation of various co-stimulatory pathways to regulate host immune responses is of therapeutic interest. In this review, we first summarize the current knowledge of T cell co-stimulatory pathways. We then focus on strategies targeting co-stimulatory pathways and their potential implications in viral vector-mediated gene therapy.

Several co-stimulatory pathways have been characterized including both activating and inhibitory pathways (Figure 1). The positive activating signals are balanced by the negative inhibitory signals to achieve optimal control of T and B cell activation. The activating pathways such as B7 and CD28, CD40 ligand (CD40L) and CD40, inducible co-stimulatory molecule (ICOS) and ICOS-ligand (ICOS-L), and OX40 (CD134) and its ligand, OX40L, are critical for the activation of T and B cells, whereas inhibitory pathways such as B7 and cytotoxic T lymphocyte antigen-4 (CTLA-4), and programmed cell death-1 (PD-1) and PD ligand (PD-L) downregulate T cell activation. Here we mainly focus on the activating co-stimulatory pathways.

Early studies in transplant setting have demonstrated that blockade of the CD28/B7 co-stimulation with CTLA-4Ig, a soluble fusion protein consisting of the extracellular domains of CTLA-4 and the constant region of the IgG1 heavy chain, could cause host T cells directed against the grafted tissue anergic, therefore leading to the long-term survival of the graft (Lenschow et al., 1993). CTLA-4Ig acts through competition with CD28 for its ligands, B7-1 and B7-2, disrupting the co-stimulation required for complete T cell activation. In addition, studies using a combination of anti-B7-1 and anti-B7-2 monoclonal antibodies (mAbs) in an allogeneic pancreatic islet transplant setting showed this approach selectively delayed CD4+ T cell infiltration into the graft, leading to inhibition of transplant rejection (Lenschow et al., 1995). Similarly, CTLA-4Ig has been used successfully to suppress T cell responses in animal models of autoimmunity (Dall’era and Davis, 2004). These studies provide evidence that blocking co-stimulatory pathway is a viable strategy in inducing T cell anergy.

A number of studies have also investigated the effect of blocking the B7–CD28 co-stimulatory pathway in viral vector-mediated gene therapy. Administration of CTLA-4Ig blocked the formation of anti-β-galactosidase antibodies following retrovirus-mediated gene transfer to the liver (Puppi et al., 2004). Similarly, treatment with CTLA-4Ig resulted in more stable transgene expression after retrovirus-mediated gene therapy in mucopolysaccharidosis I mice (Ma et al., 2007). Furthermore, potent immunosuppression has been observed with a high-affinity variant of CTLA-4Ig, LEA29Y (belatacept; Larsen et al., 2005). In a mouse model of Duchenne muscular dystrophy (DMD), adenovirus vectors carrying murine CTLA-4Ig (AdmCTLA-4Ig) was coadministered with an adenoviral vector carrying a full-length murine dystrophin cDNA (AdmDys) into skeletal muscle (Jiang et al., 2004a). Stable expression of dystrophin in skeletal muscle up to 8 weeks was achieved, in contrast to the control group without CTLA-4Ig where dystrophin expression was significantly diminished within 8 weeks. CTL response to adenoviral vector was markedly suppressed in the group receiving co-administration of AdmCTLA-4Ig at 2-week time point, and cytokine production such as IFN-γ, IL-4, and IL-6 were much lower than the control group as well. No dystrophin-specific CTL response was observed in either group during the 2-week period. Neutralizing antibody against adenoviral vector was detected in both groups, with highest levels found in the control group, indicating that CTLA-4Ig partially blocked the humoral response against adenoviral vectors.

Similar strategies employing CTLA-4Ig to block the B7–CD28 co-stimulatory pathway have been investigated in a murine model of type II collagen-induced arthritis (Ijima et al., 2001). However, in other model systems, the effect of AdmCTLA-4Ig has been modest (Laumonier et al., 2003), suggesting other strategies in combination with CTLA-4Ig may be needed to achieve successful therapeutic effect.

The critical role of the CD40L–CD40 pathway in activation of both T and B cells suggests interfering with this pathway may prevent both cellular and humoral immune responses to virus, which is critical for stable transgene expression and the repeated administration of the viral vectors. Indeed, initial studies revealed that a transient blockade of CD40 signaling with an antibody to CD40L infused at the time of vector administration resulted in stable transgene expression and diminished production of neutralizing antibodies in a murine model of adenovirus-mediated gene transfer to the liver and the lungs (Yang et al., 1996). Later, anti-mouse CD40L mAb in combination with CTLA-4Ig has been shown to further prolong adenovirus-mediated transgene expression after both primary and secondary vector delivery (Kay et al., 1995). Similarly, the combined CD40L blockade and CTLA-4Ig approach also proved effective in a murine model of retrovirus-mediated gene therapy for mucopolysaccharidosis (Ma et al., 2007). More recently, the dual blockade strategy has been shown to act synergistically to prevent antigen-specific immune responses in non-human primates, leading to stable expression of transgene and diminished neutralizing antibody production, and in some cases allowing for repeated administration of adenovirus (Haegel-Kronenberger et al., 2004).

The effectiveness of the combined antibody blockade strategy led to the evaluation for the effect of systemic administration of AdmCTLA-4Ig and AdmCD40Ig (adenoviral vector carrying murine CD40Ig) in murine models of gene delivery to the muscle (Jiang et al., 2004b) and the liver (Schowalter et al., 1997). These studies showed that systemic delivery of both AdmCTLA-4Ig and AdmCD40Ig was required to inhibit the production of anti-adenoviral vector neutralizing antibody, and that AdmCTLA-4Ig alone was insufficient. This may be due to insufficient circulating level of mCTLA-4Ig to prevent anti-adenoviral capsid protein neutralizing antibody production, and that addition of circulating levels of mCD40Ig decreases B cell activation by T cells hence leading to diminished neutralizing antibody production.

Taken together, these data lend support to the effectiveness of the combined CD40 and CD28 blockade strategy in viral vector-mediated gene therapy in animal models. However, the efficacy of this approach in human gene therapy requires further exploration in clinical trials.

T cell-dependent B cell responses are critically dependent on the interaction of ICOS with ICOS-L, which promotes germinal center formation and immunoglobulin class switching in vivo (Mcadam et al., 2001; Tafuri et al., 2001). It has been shown that anti-ICOS alone or in combination with CD40Ig or anti-CD40L can induce tolerance to islet allografts (Nanji et al., 2006) and inhibit allograft rejection in murine models of transplantation (Guillonneau et al., 2005; Taylor et al., 2005). In a murine model of non-viral mediated gene therapy for hemophilia A, blockade of the ICOS–ICOS-L pathway by anti-ICOS antibody alone was effective in inhibiting the formation of anti-factor VIII antibodies following plasmid DNA-mediated gene transfer (Peng et al., 2008). In this study, anti-ICOS treatment also resulted in elevation of Treg numbers and their suppressive activity (Peng et al., 2008). However, the efficacy of ICOS blockade has not been tested in animal models of viral vector-mediated gene transfer.

Recent advances in our understanding of the role of the innate immunity in adaptive immune responses suggest potential novel strategies for targeting the co-stimulatory pathways. The innate immune system is the first line of defense against invading pathogens through recognition of pathogen-associated molecular patterns (PAMPs) by a set of receptors called pattern recognition receptors (PRRs; Huang and Yang, 2009). Among the PRRs identified to date, the best studied is the toll-like receptor (TLR) family. Thirteen TLRs have been identified in mammals, with each recognizing a unique set of PAMPs. Most TLRs signal through the myeloid differentiating factor 88 (MyD88)-dependent pathway, initiating the downstream signaling cascade leading to the induction of proinflammatory response characterized by the production of cytokines and chemokines (Akira et al., 2006). In addition, TLR pathways are critical for the ensuing adaptive immune responses by promoting DC maturation and function through upregulation of co-stimulatory molecules (Iwasaki and Medzhitov, 2004; Huang and Yang, 2010).

Recent studies have demonstrated that the innate immune response to AAV is mediated by TLR9 (Zhu et al., 2009; Martino et al., 2011). AAV mainly activates the TLR9–MyD88 pathway in pDCs, leading to the production of type I IFNs (Zhu et al., 2009). In vivo, the TLR9–MyD88 pathway is crucial to the activation of CD8 T cell response to both the transgene product and the AAV vector, leading to the loss of transgene expression and the generation of neutralizing antibodies to transgene product and AAV (Zhu et al., 2009). Thus, the TLR9–MyD88-type I IFN pathway plays a crucial role in promoting the adaptive immune responses to AAV.

Stimulation of TLR9 with its ligand, CpG-DNA leads to upregulation of surface expression of co-stimulatory molecules, such as CD40, B7-1/B7-2, and MHC class II on cDCs and macrophages (Hemmi et al., 2003). Therefore, DC maturation and Th1-like cytokine induced by TLR9 stimulation result in efficient and robust activation of Th1 and CD8 cytotoxic T lymphocytes (CTL; Lipford et al., 2000; Horkheimer et al., 2009). Type I IFNs, in addition to their direct anti-viral effects (Vaidya and Cheng, 2003), have also been shown to induce DC maturation by upregulating the expression of co-stimulatory molecules, which in turn promotes T cell responses (Hoebe et al., 2003). DCs matured by type I IFNs promote cross-priming of virus-specific CD8 T cells (Le Bon et al., 2003). Type I IFN can also promote effector function of virus-specific T cells (Cousens et al., 1999), and enhance the survival of activated T cells (Marrack et al., 1999). Furthermore, type I IFN signaling on DCs is important for the production of virus-specific IgM, whereas the generation of protective neutralizing antibodies to adenovirus critically depends on the type I IFN signaling on both CD4 T cells and B cells (Zhu et al., 2007).

A critical role for the TLR9 signaling pathway in the upregulation of co-stimulatory molecules and the induction of adaptive immune response to AAV suggests that strategies targeted to interfere with this signaling pathway may diminish T cell responses to the AAV vector and the transgene product, leading to prolonged transgene expression and reduction in inflammation, improving the safety and efficacy of AAV vectors for gene therapy in humans. Furthermore, blockade of the TLR9 signaling pathway will also likely inhibit antibody responses to the transgene product as well as the AAV vector.

One way to interfere with the TLR9–MyD88-type I IFN signaling pathway is to use antagonists specific to TLR9 such as inhibitory ODN to block the TLR9–MyD88 pathway (Zhu et al., 2009; Martino et al., 2011). Alternatively, blockade of type I IFNs by administering neutralizing antibodies to IFN-α and IFN-β may also prove effective as type I IFNs are required for adaptive immune responses to AAV (Zhu et al., 2009).

In this review, we have summarized the role of various T cell co-stimulatory pathways in the activation of both T and B cells and strategies targeting these pathways to improve the outcome of viral vector-mediated gene therapy. The majority of the studies so far are focused on targeting the B7–CD28 and the CD40L–CD40 pathways. It may be beneficial to investigate the effect of blocking other positive co-stimulatory pathways such as ICOS and ICOS-L, and OX40 and OX40L. In addition, strategies for stimulating the negative co-stimulatory pathways such as PD-1 and PD-L may prove effective. Given that the combined blockade of CD40 and CD28 pathways is more effective than the blockade of a single pathway alone, future studies should focus on targeting multiple pathways including both the positive and negative co-stimulatory pathways. In this regard, since the innate immunity is critical in regulating multiple co-stimulatory pathways, strategies targeting innate immune pathways may warrant further investigation. Ultimately, the utility of these strategies in improving the outcome of gene therapy with viral vectors will need to be tested in human clinical trials.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.